# Conflicts:
#	ROADMAP.md
#	docs/function-parity.md
This commit is contained in:
Brandon Thomas
2026-10-24 02:10:42 -04:00
5 changed files with 147 additions and 18 deletions
+3 -1
View File
@@ -2,7 +2,7 @@
**Stage: alpha** · next: beta, ~25 points of "ready for real work" and ~160–240 h away
> **Last reviewed:** 2026-10-11 · **Last updated:** 2026-10-23 · **Change:** minor (INDEX with one index on a 2-D reference is #REF! as in Excel; earlier: Page Setup paper sizes: JIS B4/B5/B6, the Japanese postcards and envelopes, and A6; long flat operator chains up to Excel's formula limit) · **Target:** Microsoft Excel (Microsoft 365)
> **Last reviewed:** 2026-10-11 · **Last updated:** 2026-10-23 · **Change:** minor (accounting negatives such as `$(1,234)` read as numbers; INDEX with one index on a 2-D reference is #REF! as in Excel; earlier: Page Setup paper sizes: JIS B4/B5/B6, the Japanese postcards and envelopes, and A6; long flat operator chains up to Excel's formula limit) · **Target:** Microsoft Excel (Microsoft 365)
GridCraft aims at full Microsoft Excel parity, and at being better in places: faster, open (XLSX
is the native format), fully scriptable by agents (MCP, control channel, CLI), and on the web.
@@ -135,6 +135,7 @@ Ranked; estimates in Opus 5.5 agent-hours. Detail and the full beta-gate list in
| Date | What landed |
|---|---|
| 2026-10-23 | Accounting negatives in text: `$(1,234)`, `$ (5)` and `(5)%` read as negative numbers when typed, coerced by arithmetic or passed to VALUE, like Excel 365; `(-5)`, `((5))`, `$5%` and dates or times in parentheses stay text |
| 2026-10-23 | Calc: `INDEX(A1:C3,2)` (one index on a reference of several rows and columns) is `#REF!` as in Excel, instead of the whole row; `INDEX(A1:C3,2,)` still gives the row |
| 2026-10-22 | Formula hardening: digits too many for a finite number are text, so `"1 <400 digits>/<400 digits>"` no longer coerces to NaN; a NaN column index in VLOOKUP, HLOOKUP and the D-functions is `#VALUE!` instead of an index underflow; REGEXREPLACE stops building a result as soon as it can only be too long (a worst case went from 1 GB to 7 MB) (#2163, #2162) |
| 2026-10-21 | Page Setup paper sizes match Japanese Excel: `B4`/`B5`/`B6` are the JIS sizes (ISO B4/B5/B6 stay reachable by name), so a Japanese B5 worksheet prints at 182×257 mm; Japanese Postcard (100×148 mm), Double Postcard (148×200 mm), Envelope #3/#4 (120×235/90×205 mm), Kaku #2/#3 (240×332/216×277 mm) and A6 are added, both sizes reach the PDF page (`file.exportPdf`/`file.print` MediaBox) and the size survives an XLSX round-trip; the Page Layout Size menu lists them. `B4`/`B5` used to be the ISO sizes, which is changed here |
@@ -223,6 +224,7 @@ Ranked; estimates in Opus 5.5 agent-hours. Detail and the full beta-gate list in
| Date | Change | Summary |
|---|---|---|
| 2026-10-23 | minor | Progress log: accounting negatives in number text |
| 2026-10-23 | minor | Progress log: INDEX with one index on a 2-D reference is #REF! |
| 2026-10-22 | trivial | Progress log: formula hardening (#2163, #2162) |
| 2026-10-21 | minor | Page Setup paper sizes: JIS B4/B5/B6 (ISO by name), Japanese postcards and envelopes, A6; PDF page size follows, XLSX round-trips |
+15
View File
@@ -427,6 +427,21 @@ fn lambda_helpers_check_shapes_and_parameters() {
assert_eq!(t.num("B7"), 2.0);
}
#[test]
fn accounting_negative_text_coerces() {
let mut t = T::new();
t.set("A1", "'$(1,234)");
t.set("B1", "=A1+0");
t.set("C1", "=VALUE(A1)");
t.set("D1", "=--\"$ (5)\"");
t.set("E1", "=\"(5)%\"*100");
t.set("F1", "=\"(-5)\"+0");
t.set("G1", "$(1,234)");
assert_eq!((t.num("B1"), t.num("C1"), t.num("D1"), t.num("E1")), (-1234.0, -1234.0, -5.0, -5.0));
assert_eq!(t.get("F1"), Value::Error(CellError::Value));
assert_eq!(t.num("G1"), -1234.0);
}
#[test]
fn operators() {
let mut t = T::new();
+120 -16
View File
@@ -61,16 +61,16 @@ pub fn number_stored_as_text(s: &str) -> Option<f64> {
}
fn parse_numeric(t: &str, sys: DateSystem) -> Option<(f64, Option<&'static str>)> {
// Parenthesised negatives: (100) = -100. Each level negates, so ((100)) = 100. Stripped in a
// loop, not by recursion: runtime text (e.g. from REPT) can nest thousands deep (#2180).
let mut t = t;
let mut neg = false;
while let Some(inner) = t.strip_prefix('(').and_then(|x| x.strip_suffix(')')) {
t = inner.trim();
neg = !neg;
// Parenthesised negatives: (100) = -100, and the accounting forms `$(100)` and `(5)%`.
if let Some(inner) = parenthesized(t) {
// Excel reads no sign and no nested parentheses inside: `(-5)`, `($-5)`, `((5))`.
if inner.strip_prefix('$').map_or(inner.as_str(), str::trim_start).starts_with(['-', '+', '(']) {
return None;
}
let r = parse_plain(&inner).or_else(|| parse_fraction(&inner).map(|n| (n, Some("# ?/?"))));
return r.map(|(n, f)| (-n, f));
}
let (n, f) = parse_unparenthesised(t, sys)?;
Some((if neg { -n } else { n }, f))
parse_unparenthesised(t, sys)
}
fn parse_unparenthesised(t: &str, sys: DateSystem) -> Option<(f64, Option<&'static str>)> {
@@ -89,6 +89,22 @@ fn parse_unparenthesised(t: &str, sys: DateSystem) -> Option<(f64, Option<&'stat
None
}
/// The number inside accounting parentheses, as plain text: `(5)` gives `5`. A dollar sign may
/// stand before them (`$(5)` reads as `($5)`) and a percent sign after them (`(5)%` reads as
/// `(5%)`), with spaces around either. The caller reads only a plain number or a fraction from it.
fn parenthesized(t: &str) -> Option<String> {
let (currency, rest) = match t.strip_prefix('$') {
Some(r) => ("$", r.trim_start()),
None => ("", t),
};
let rest = rest.strip_prefix('(')?;
let (body, percent) = match rest.strip_suffix(')') {
Some(b) => (b, ""),
None => (rest.strip_suffix('%')?.trim_end().strip_suffix(')')?, "%"),
};
Some(format!("{currency}{}{percent}", body.trim()))
}
/// Signs, currency, thousands separators, percent, exponent.
fn parse_plain(t: &str) -> Option<(f64, Option<&'static str>)> {
let mut s = t;
@@ -113,7 +129,8 @@ fn parse_plain(t: &str) -> Option<(f64, Option<&'static str>)> {
percent = true;
s = r.trim_end();
}
if s.is_empty() {
// Excel reads `$5` and `5%` but not `$5%`.
if s.is_empty() || currency && percent {
return None;
}
let has_sep = s.contains(',');
@@ -422,20 +439,107 @@ mod tests {
assert_eq!(p(".5").value, Value::Number(0.5));
}
/// Observed in Excel 365 (en-US): typed entry, VALUE and `--"text"` agree.
#[test]
fn accounting_parentheses() {
let accepted: [(&str, f64, Option<&str>); 22] = [
("$(5)", -5.0, Some("\"$\"#,##0")),
("$ (5)", -5.0, Some("\"$\"#,##0")),
("$( 5 )", -5.0, Some("\"$\"#,##0")),
("$ ( 1,234 )", -1234.0, Some("\"$\"#,##0")),
("$(1,234)", -1234.0, Some("\"$\"#,##0")),
("$(1,000.50)", -1000.5, Some("\"$\"#,##0.00")),
("$(1,234.)", -1234.0, Some("\"$\"#,##0.00")),
("$(.5)", -0.5, Some("\"$\"#,##0.00")),
("$(0)", 0.0, Some("\"$\"#,##0")),
(" $(5) ", -5.0, Some("\"$\"#,##0")),
("($5)", -5.0, Some("\"$\"#,##0")),
("( $ 1,234 )", -1234.0, Some("\"$\"#,##0")),
("(5)", -5.0, None),
("( 5 )", -5.0, None),
("(1,234.5)", -1234.5, Some("#,##0.00")),
("(1e-3)", -0.001, Some("0.00E+00")),
("(5)%", -0.05, Some("0%")),
("( 5 ) %", -0.05, Some("0%")),
("(5%)", -0.05, Some("0%")),
("(5.)%", -0.05, Some("0.00%")),
("(1,234)%", -12.34, Some("0%")),
("(1 1/2)", -1.5, Some("# ?/?")),
];
for (s, n, f) in accepted {
assert_eq!(p(s), Parsed { value: Value::Number(n), format: f }, "{s}");
assert_eq!(parse_number_text(s), Some(n), "{s}");
}
for s in [
"$5%",
"$(5%)",
"$(5)%",
"($5%)",
"($5)%",
"$ ( 5 ) %",
"(5)%%",
"(5%)%",
"((5))",
"(-5)",
"(+5)",
"($-5)",
"($ -5)",
"($+5)",
"(-$5)",
"$(-5)",
"$ (-5)",
"$(+5)",
"-(5)",
"-$(5)",
"+$(5)",
"$+(5)",
"$$(5)",
"$ $(5)",
"$((5))",
"($(5))",
"(-1 1/2)",
"()",
"$()",
"(5)$",
"$(5)$",
"(5)(5)",
"$(5)0",
"(5)e3",
"(5)-",
"(5-)",
"$(5-)",
"$(5",
"$5)",
"(5",
"$(1,23)",
"$(,5)",
"(5e)",
"(1/2)",
"$(1/2)",
"(3/4/2026)",
"$(3/4/2026)",
"(10:30)",
"$(10:30)",
] {
assert_eq!(p(s).value, Value::text(s), "{s}");
assert_eq!(parse_number_text(s), None, "{s}");
}
}
#[test]
fn deeply_nested_parentheses_do_not_overflow() {
// #2180: coercion used to recurse once per parenthesis and overflow the stack.
// #2180: coercion used to recurse once per parenthesis and overflow the stack. Excel reads
// one pair only, so nesting is text now, and it is rejected without recursing.
let depth = 100_000;
let s = format!("{}1{}", "(".repeat(depth), ")".repeat(depth));
assert_eq!(parse_number_text(&s), Some(1.0));
assert_eq!(parse_number_text(&s), None);
assert_eq!(p(&s).value, Value::text(s.as_str()));
let odd = format!("{}1{}", "(".repeat(depth + 1), ")".repeat(depth + 1));
assert_eq!(parse_number_text(&odd), Some(-1.0));
assert_eq!(parse_number_text(&odd), None);
let unbalanced = format!("{}1{}", "(".repeat(depth), ")".repeat(depth - 1));
assert_eq!(parse_number_text(&unbalanced), None);
assert_eq!(p(&s).value, Value::Number(1.0));
// Existing results are unchanged.
// A single pair still reads.
assert_eq!(parse_number_text("(100)"), Some(-100.0));
assert_eq!(parse_number_text("((100))"), Some(100.0));
assert_eq!(parse_number_text("( $1,200.50 )"), Some(-1200.5));
assert_eq!(parse_number_text("()"), None);
}
+7
View File
@@ -1211,6 +1211,13 @@ mod tests {
close(ev("VALUE", vec![t("12:00")]), 0.5);
close(ev("VALUE", vec![t("2020-01-01")]), 43831.0);
is_err(ev("VALUE", vec![t("abc")]), CellError::Value);
// Accounting negatives, as Excel 365 reads them.
close(ev("VALUE", vec![t("$(5)")]), -5.0);
close(ev("VALUE", vec![t("$ (1,000.50)")]), -1000.5);
close(ev("VALUE", vec![t("(5)%")]), -0.05);
for s in ["$(5)%", "$(5%)", "$5%", "(-5)", "((5))", "$(-5)", "(1/2)"] {
is_err(ev("VALUE", vec![t(s)]), CellError::Value);
}
close(ev("NUMBERVALUE", vec![t("2.500,27"), t(","), t(".")]), 2500.27);
close(ev("NUMBERVALUE", vec![t("3.5%")]), 0.035);
close(ev("NUMBERVALUE", vec![t(" 1 000 ")]), 1000.0);
+2 -1
View File
@@ -1,6 +1,6 @@
# Function parity: worksheet functions and formulas
> **Last reviewed:** 2026-10-15 · **Last updated:** 2026-10-23 · **Change:** minor (INDEX with one index on a 2-D reference is #REF! as in Excel; earlier: SUBTOTAL and AGGREGATE cap the cells they visit, #2175) · **Target:** Microsoft Excel (Microsoft 365), Excel for Mac 16.113.4
> **Last reviewed:** 2026-10-15 · **Last updated:** 2026-10-23 · **Change:** minor (number text with accounting parentheses coerces like Excel 365; INDEX with one index on a 2-D reference is #REF! as in Excel; earlier: SUBTOTAL and AGGREGATE cap the cells they visit, #2175) · **Target:** Microsoft Excel (Microsoft 365), Excel for Mac 16.113.4
Excel's worksheet functions, category by category, against GridCraft's function library. Part of
[target-app-parity.md](target-app-parity.md); gaps feed [gaps.md](gaps.md).
@@ -123,6 +123,7 @@ workbooks; results checked in as data, not Excel files) and fixing what it finds
| Date | Change | Summary |
|---|---|---|
| 2026-10-23 | minor | Number text coercion (VALUE, arithmetic, typed entry): a dollar sign before accounting parentheses (`$(1,234)`, `$ (5)`) and a percent sign after them (`(5)%`) read as negatives; `(-5)`, `((5))`, `($-5)`, `$5%`, `$(5)%` and dates or times in parentheses are `#VALUE!`, as observed in Excel 365 |
| 2026-10-23 | minor | INDEX with only a row index on a reference of several rows and columns (`INDEX(A1:C3,2)`, `INDEX(A1:C3,0)`) is `#REF!`, as observed in Excel, so it also fails as a range endpoint and inside other functions; an empty column argument (`INDEX(A1:C3,2,)`) still gives the row. Checked that a column array with one index stays a 1×1 array (`TYPE` 64), which is what Excel returns |
| 2026-10-21 | minor | SUBTOTAL and AGGREGATE charge every argument against the evaluator's 16-million-cell limit before scanning; a value in XFD1048576 no longer makes `SUBTOTAL(9,A:XFD)` read the whole sheet (#2175) |
| 2026-10-19 | minor | LET and LAMBDA names compare case-insensitively in any script, as in Excel: `=LET(dług,100,DŁUG*2)` is 200 (was #NAME?); also LET-bound lambdas called or passed by name (`żar(2)`, `MAP(…,Λ)`), `ISOMITTED`, and formula-bar alias translation (one shared `gridcraft_formula::same_name`, which matches `ß` with `ẞ` and keeps dotless `ı` apart from `i`) |